OpenDUck improvements

[simonbray]

Suggestions from the meeting today:

• Single central repo, on GitHub, with software versions
• Conda env resolution: Update dependencies, make pinning more flexible if possible
• Add scripts to setup.py as entry points
• Publish on conda and PyPI
• getWqbValues.py could be included as a helper script, not written by the library
• Allow amber script to be outputted as a simple bash script, as well as slurm and SGE
• Parameterisation of openduck scripts? Cannot judge what is sensible here, what should be hardcoded and what not.
• • (from Dani) array value in slurm script
• Random seed option in scripts; add reproducible test to repo
• • Solvation - is this deterministic?
• (from Dani) a script for analysis, to create a summary table of values / results
• (from Dani) add final wqb value to sd-file
• Filter scripts directory, remove ones which are not used
• (from Dani) not always clear how to set threshold min_wqb values for a new system, needs experimentation, create a utility script to make this smoother
• (from Xavi) something equivalent to rbhtfinder in rdock, to suggest for an initial run what compounds are promising to continue with
• Error handling…

ping @AlvaroSmorras @DAlvGar @xbarril

[AlvaroSmorras]

Sorry about that,
the solvation seems deterministic in PDBFixer (https://github.com/openmm/pdbfixer/blob/db2886903fe835919695c465fd20a9ae3b2a03cd/pdbfixer/pdbfixer.py#L1051) and there's no seed argument in the function either.

[simonbray]

@AlvaroSmorras when I try to run the same command twice, the positions of the water molecules recorded in the inpcrd file are different. So I assume it cannot be deterministic. But I admit, I don't see anything in that code which suggests randomness. 😕

[AlvaroSmorras]

@AlvaroSmorras when I try to run the same command twice, the positions of the water molecules recorded in the inpcrd file are different. So I assume it cannot be deterministic. But I admit, I don't see anything in that code which suggests randomness. confused

Hmh... the PDBFixer uses the openmm.app Modeller class with the addSolvent function. But it does not have any seed option either. I added the seed option for the equilibration,Md and SMDs, however.

[AlvaroSmorras]

Comment from @PatriciaBlanco, sort the report table by the system name. It can be easily done by sorting the output df here: https://github.com/CBDD/openduck/blob/14767e5c46d7800b337a36f8be60351090be9f1c/duck/utils/analysis_and_report.py#L118C12-L118C14